4.1.3. Real-Time Hub and Alerts
💡 First Principle: Real-Time Hub provides visibility into streaming data flows, and alerts enable proactive notification when thresholds are breached. Instead of discovering problems during morning review, you know immediately when something goes wrong.
Scenario: An eventstream processes payment transactions. If the error rate exceeds 1%, the operations team must be notified immediately—not discovered during morning review.
Configuring Alerts
- Navigate to Real-Time Hub
- Select event source or stream
- Configure alert conditions
- Specify notification method (email, Teams)
Alert Types
| Alert Type | Trigger | Use Case |
|---|---|---|
| Workspace Events | Pipeline failed, refresh failed | Failure notification |
| Metric Threshold | Error rate > X%, Latency > Y ms | Performance degradation |
| Data Condition | Specific value in stream | Business rule violation |
⚠️ Exam Trap: Configure alerts for performance degradation, not just failures. Latency increases and throughput drops often precede failures—catching them early prevents outages.
What else Real-Time hub gives you:
- A tenant-wide catalog of streams: every eventstream output and KQL table you can access is listed automatically. Look here to discover existing real-time sources before building a new one; the OneLake catalog is for data items, not streams.
- Fabric events you can alert on: Job events (a job such as a pipeline run, notebook run or semantic model refresh being created, succeeding or failing), Workspace item events (items created, changed or deleted), OneLake events (files and folders created or deleted), and Capacity overview events (capacity utilization and throttling thresholds). Azure Blob Storage events are available too.
- Fabric Activator (formerly Data Activator) is the rule engine behind these alerts: it watches a stream or event, evaluates a condition, and acts by sending an email or Teams message or running a pipeline or notebook.
For a failed pipeline run you can also alert from inside the pipeline by routing the failing activity's On fail path to an Office 365 Outlook or Teams activity (see 4.2.1), then review what happened in Monitor hub (4.1.1).
